Open Collections

Abstract

A large amount of psychological research is devoted to the investigation of the psychological impact of emotionally relevant stimuli. A key problem that characterizes this field of study is the difficulty of attributing a psychological effect to affective content itself, rather than to confounded low-level stimulus properties. When conducting this research with visual stimuli, the alternative explanation of confounded visual properties is often tested by disrupting the perception of higher-order visual information while maintaining low-level visual properties. The texform method (Long, Konkle, Cohen, & Alvarez, 2016) is a recent general-purpose way to accomplish this goal. However, this method is limited in important ways: (1) it cannot be controlled in a fine-grained manner to allow for different levels of visual organization, (2) it routinely contains noticeable visual artifacts which induce straight lines and structure that were not part of the input stimulus, and (3) it is very slow to execute, making it difficult to implement for studies that use a large number of visual stimuli. The present set of studies develop and assess two new methods that correct for these limitations in the texform method. In the course of developing these methods, I use them to address an example affective science research question: whether ecological threats are recognizable from degraded information, as would be expected from an evolved fear module (Öhman & Mineka, 2001). Through studies with human participants, both new methods are validated as disrupting visual categorization in a dose-dependent manner. However, one of these methods, the seeded texform method, is established as superior via subjective assessment and computational measures of image structure. This method is shown (1) to affect image structure in a predictable, monotonic manner, (2) to affect explicit classifiability of objects, and (3) to affect implicit classifiability, as measured in the fast periodic visual stimulation (FPVS) oddball paradigm. The seeded texform method is recommended as a broadly applicable method for fine control over the higher-order visual properties of a visual stimulus. Empirical tests of the affective influence on behavioural and electrophysiological measures are interpreted as inconsistent with the hypothesized “perceptual generalization” from the fear module theory.